libcamera: rpi: pipeline_base: Cache sensor format

The format to be applied on the sensor is selected by two criteria: the
desired output size and the bit depth. As the selection depends on the
presence of a RAW stream and the streams configuration is handled in
validate() there is no need to re-compute the format in configure().

Centralize the computation of the sensor format in validate() and remove
it from configure().

Signed-off-by: Jacopo Mondi <jacopo.mondi@ideasonboard.com>
Reviewed-by: Naushir Patuck <naush@raspberrypi.com>

1 files changed, 2 insertions, 0 deletions

diff --git a/src/libcamera/pipeline/rpi/common/pipeline_base.h b/src/libcamera/pipeline/rpi/common/pipeline_base.h
index 2eda3cd8..a139c98a 100644
--- a/src/libcamera/pipeline/rpi/common/pipeline_base.h
+++ b/src/libcamera/pipeline/rpi/common/pipeline_base.h
@@ -262,6 +262,8 @@ public:
 
 	/* Cache the combinedTransform_ that will be applied to the sensor */
 	Transform combinedTransform_;
+	/* The sensor format computed in validate() */
+	V4L2SubdeviceFormat sensorFormat_;
 
 private:
 	const CameraData *data_;